Integration of the Grain-Size Statistic Data Populations and the Facies Model 180

Fig. 4.1 presents a first attempt at integrating grain-size data into the facies model. None of the populations other than 1 is correlated with distinct facies types (Ll-3). Nos does the first order cyclicity, which is linked to depositional energy, correlate with mean grain-sizes of the samples (Fig. 4.37). To overcome the limitations of the sequence stratigraphic facies model, the facies ase broken down to their basic units, assuming that the sparse sampling was fine enough to resolve single process-related sediment volumes. Depositional processes thought to act during drift build-up and maintenance are:

1) Hemipelagic settling dusing times of slow and low depositional energy, often associated with intense bioturbation and an enrichment of IRD. The intensity of bioturbation, amount of ice-rafted debris, and extent of hemipelagic bed tops depend on the recurrence interval of turbidite events. Facies association TE3-H.

2) Deposition and partial erosion by means of bottom contour currents, causing winnowing of already deposited sediment. Resulting in coarse skewed grain-size frequency distribution curves. This process is highly variable in intensity at longer time periods and may affect a large spectrum of grain-sizes. At the same time this process influences all short-living depositional processes. The base units bof rapid depositional sequences (turbidite-, debris style sedimentation) may be not affected because they are quickly buried. Erosion and redeposition from bottom currents especially affects topic 1 processes and to a lesser extent processes described under topic 3. Facies association C, H to a lesser extent E-D.

CHAPTER 4: The West Anturcfic Contitzental Rise

Schematic lithologic columns with first-order cycles

Fig. 4.1. Simplified schematic lithostratigraphy of Site 1095 showing dominant lithology, intensity of bioturbation, and distribution of turbidite Facies L l , L2, and L3. Arrows show broad trends in frequency of sand and silt larninations and facies types, attributed to long-tei'm "first-order" cycles in lithostratigraphic Unit I1 according to the shipboard model. Grain-size populations 1-3 as defined within this chapter are scaled next to the descriptive units. The length of the bars refers to the relative mean grain-size of the populations.

CHAPTER 4: The West Antorctic Continental Rise

3) Lower energy turbidites with no or minor basal erosion. These are attributed to slope failure under interglacial conditions (no grounded ice at the shelf edge). Under interglacial conditions the slope is fed with glacial meltwater debris, minor IRD and pelagic material. Entrainment to the rise occurs occasionally via distal overbank deposits and feeding of a nepheloid layer (Pudsey and Camerlenghi, 1998; Rebesco et al., 1996) that is then moved by bottom currents and acts as a distributor reaching remoter Parts of the drift bodies. Facies association D-E.

4) Higher energy turbidites and debris flows. These are most likely to occur under conditions in which shelf ice is grounded and reaches to the shelf edge, periodically dumping eroded inland and shelf topset material. In addition, massive IRD input to the upper and lower slope reduces transport distances, allowing coarse and unsorted material to reach the drifts. The recurrence interval of successive turbidite events during shelf ice advances is probably short, which reduces the time for undisturbed hemipelagic accumulation. Facies association TC-TD.

4.6.2.1 Grain-Size Population 3

Population 3 has the lowest average mean (Fig. 4.1) but is the most variable population with respect to its skewness, mean and standard deviation (sorting) values (Fig.

4.1). The depositional process responsible for this chasacteristic must be continuously scalable in depositional energy, but at the same time retains distinctly better sorting, even for populations that show comparable mean values than does ~ o ~ u l a t i o n 2 (Fig. 4.1 C).

Compasisons with core descriptions and the simplified lithological column (Fig. 4.1) link this population to bioturbated intervals. Bioturbated intervals are found in the sequence tops of all proposed facies. Sequence tops (depositional process 1) have the slowest accumulation rates and are subsequently longest exposed to the action of bottom currents and IRD input.

However, the limitation of our study to the sortable silt fraction (10-63 um) has effectively avoided the classical IRD spectrum (sand fraction). The fact that the sum frequency distribution curve of Population 3 is positively skewed (tail in the fines) is interpreted by Pudsey and Camerlenghi (1998) as an indicator of only minor current winnowing (removal of the fines) and, hence, as a negligible contribution of bottom currents to the depositional process. Instead the good correlation of opal content (Fig. 4.1) and skewness (Fig. 4.3 B) rather suggests that the constant influx of comparably coarse organic opal particles OS

fragments accumulated at sequence tops mask the true nature of the frequency distribution of this population (see also the good correlation of sand and diatom content in the paper of

182

CHAPTER 4: The West Antarcfic Continental Rise

Pudsey, in press). Another argument for the importance of bottom cusrents for the definition of Population 3 is the comparably better sosting of this population, since contourite currents are known to increase sosting of sediments (Folk, 1974).

Population 3 is therefore associated with facies subdivision TE3-H and C, and describes low energy hemipelagic, bioturbated sediments under the influence of fluctuating bottom currents. Together with the skewness, kustosis and opal data the Population 3 density curve may as well be an anti-cosrelated sea-ice indicator. Accepting that the present link between atmospheric wind (polar easterlies) and bottom currents (westward flow of AABW) being true for the past, Population (especially the density curve, Fig. 4.1) is also positively cosrelated to bottom-current strength.

4.6.2.2 Grain-Size Population 2

Population 2 shows the smallest variability in standard deviation and mean grain size and has the same kind of tailing toward smaller skewness values as does Population 3 (Fig. 4.1). The depositional process responsible for this characteristic must be closely defined regarding energy and sorting, but at the Same time must allow for continuous degrees of winnowing as expressed by the tailing toward smaller skewness values. Based On this observation, Population 2 is associated with facies subdivisions D-E and processes described under topic 3.

Supporting this classification as ,,interglacial turbidite" is the continuous decline in mean grain size of Population 2 upsection, probably addressed to the build-up of the drift above the turbidite channels, making it more difficult to deposit coarser material On the drift crest. The large variability in skewness points to some bottom-cusrent influence, especially on the finer grained members of this data population. The sirnilarities in mean grain size to Population 1 may indicate that past of the deposited material of the ,,interglacial turbidites" is reworked organic material deposited On the slope. Other constituents probably contain distal meltwater debris from the Peninsula transported to the slope by shelf currents. From a mass flux point of view, sediment supply thsough mechanism 3 (Population 2) may represent the "cruising gear"

supplying continuously small volumes throughout the drift evolution. The observed exponential decrease in mass flux with time (Fig. 4.3) may be related to the decrease of mean grain sizes deposited by this process in an upsection direction with time.

4.6.2.3 Grain-Size Population 1

Population 1 is the coarsest and most unsorted unit with negligible tailing along the skewness axis (Fig. 4.1). From Fig. 4.2 B it is evident that the average distribution curve of Population 3

CHAPTER 4 : The West Antarctic Continentul Rise

is perfectly symmetrical. This would indicate the presence of no or only the subtle impact of bottom current. The depositional process responsible for this characteristic must be fast and high in depositional energy. Population 3 is therefore associated with facies subdivisions TD- TC and processes described under topic 4. Supporting this classification as ,,glacial turbidite"

is the good correlation of Population 1 with intervals of > 63 pm (Fig. 4.4) and IRD time intervals (<250 pm) of the rise (Cowan, in press). Since it has been shown that the fine fraction data effectively avoid direct IRD influence (chapter 4.3) the question as to whether the ice-related Population 1 is strengest during a glacial period or occurs at the transition from a glacial to an interglacial (see Pudsey and Camerlenghi, 1998) is of minor importance (sample resolution is not fine enough to resolve single glacial-interglacial variations).

Population l is seen here as a largely IRD-independent Parameter that indicates advances of continental ice to the shelf edge with only short delay times of the sediment on the slope.

Population 1 works as an ice indicator also in the lower Part of the core where classical IRD definitions, e.g. by Cowan (in press) would fail.

Comparisons of the density curve calculated from Population 1 data (chapter 4.3) with a new global ice volume curve by Lear et al. (2000) shows, that the Messinian and the lower Pliocene have been times of reduced ice volume also reflected in the Population 1 data of the Antarctic Peninsula (Fig. 4.1). Estimates of how dramatic an early Pliocene deglaciation might have been are drawn based on assigning 100% global ice volume to the Pleistocene (Lear et al., 2000). Following the Lear study, global ice volume has been reduced by -80%

throughout the early Pliocene. When taking the Population 1 data is taken into consideration quantitatively, ice volume on the Antarctic Peninsula was reduced by more than 70%

(assigning 100% to the maximum values occurring in the Pliocene). Starting from total different perspectives the findings of this study and that of Lear et al. (2000) are in excellent agreement. A deglaciation in the Messinian and early Pliocene is further supported by the eustatic data of Haq et al. (1987) which shows three successive sea-level highs sta~ting in the Messinian and ending in the lowermost upper Pliocene.

These findings, of Course, do not define the spatial extent of deglaciation in Antarctica.

But a 70-80% reduction in ice volume should require at least ice-free low lands around Antarctica and possible time periods with no ice on the more exposed Antarctic Peninsula (see also remarks by Barrett, 2001).

CHAPTER 4: The West Antarctic Continental Rise data. The ice volume is calculated by removing the temperature effect from benthic forarninifera 6^0 data using a Ca-Mg thermometer. Dark gray hoxes indicate times of major ice growth. (Part (A) of this figure is modified from Lear et al., 2000).

CHAPTER 4: The West Aniarctic Continental Rise

4.7 Summary: The Rise as a Recorder of Ice and Currents over Time